The Farnese letter | Simon Klee

In April 1542 a letter left Rome for the courtroom of Charles V in Spain. On its
first web page the Italian stops in the course of a line and digits start:

Figure 1. The opening of the cipher, f. 70r. Archivio Apostolico Vaticano
(AAV), Segr. Stato, Spagna 1A, {photograph} equipped by way of
DECODE record 92.
Detail enlarged from the {photograph}.

Read with the important thing recovered beneath, the primary ten digits group like this:

73  4  57  4  9  03  5
 d  o   p  o  ·   l  a

Dopo la, “after the”. The 9 stands for nothing: it’s a null. The
problem is deciding the place every code ends. 73 is d, however 7 3 can also be
two letters, n m. Even with the important thing, a run of digits might be learn greater than
a technique.

Two issues comply with: discovering the important thing with out realizing the place one code ends
and the following begins, after which, as soon as a key exists, telling the author’s
phrases from a believable guess.

The letter

The sender was Cardinal Alessandro Farnese, grandson of Paul III and, at
twenty-one, head of the papal secretariat. The recipient was Giovanni Poggio,
bishop of Tropea and nuncio, the Pope’s ambassador, on the Emperor’s
courtroom. ^{The letter opens in odd Italian: Poggio may have heard
by phrase of mouth from Giovanni Ricci da Montepulciano about la materia della
tempo, the peace between the Emperor and France, so Farnese is not going to repeat
it. Montepulciano had returned from Spain in February with the Emperor’s
phrases and had left Rome for Spain once more in late March.}

Montepulciano’s identify can also be on a cipher key. Aloys Meister’s 1906 research of
papal cryptography prints, among the many previous ciphers of the papal secretariat, a
key headed Cifra ultima con Mons. Poggio mandata per il Montepulciano: the
newest cipher with Poggio, despatched by Montepulciano. Meister dates it 1538–42.

The letter fills eight pages on 4 leaves, 70r to 73v within the archive’s
numbering: r for the entrance of a leaf, v for the again. Page 70r is
cleartext for many of its size. Then the digits take over within the center
of a line and fill the entire of pages 70v to 72r. On 72v they encompass a
cleartext passage about church trade and information from Ancona, “recent and
from an excellent supply”. The cipher ends on 73r, and the remainder of that web page is
clear. Page 73v is evident and dated Da Roma alli X[?] di Aprile 1542: the
day is a Roman numeral starting with X whose remaining strokes lie underneath a
flourish, nonetheless unread. What is in clear is the body; the directions are
in cipher.

Figure 2. Ff. 70r and 70v. AAV, Segr. Stato, Spagna 1A, by way of
DECODE record 92, public
previews. The change from prose to digits comes close to the foot of 70r.
The cipher continues throughout the following web page.

Ludwig Cardauns cited this quantity in his 1912 assortment of the nuncios’
studies, however his chosen correspondence doesn’t embody this letter.
^{In July 2019 George Lasry set the textual content as Part 5 of the Vatican Challenge
on MysteryTwister, a web site of cipher puzzles: key unknown, plaintext unknown,
language most likely Italian.
In their research, revealed on-line in 2020, Lasry, Beáta Megyesi, and Nils
Kopal nonetheless listed the Spagna 1A materials as unsolved. Their digit-frequency
evaluation positioned it in a separate cluster. They in contrast one other a part of
the gathering, IA-1, with Meister’s Poggio key, with out success, and thought
this letter, IA-2, used a special key. On 16 September 2026,
the problem web page nonetheless displayed zero solves.}

The search started with a public transcription, now distributed with the
DECODE file and the problem. Its header credit EHum, dates the
transcription 9 January 2016, and information about six and a half hours’ work
on the eight pages. The transcription comprises 6,577 cipher digit positions:
44 recorded as ?, unreadable, and 207 with a mark above them, largely dots.
There is not any constant phrase division. The full ciphertext
is reproduced as textual content, with its web page and line breaks. The counts and searches
beneath used this public transcription. I checked it towards the bigger
images later.

What the counts present

In the general public transcription, digit 7 makes up 17.5 p.c of the textual content and
1 solely 3.1. A couple of pairs are way more frequent than probability: 80 happens 349
instances, 57 317, 27 263, 03 258, 73 220. Doubled digits are practically
absent: 00 six instances, the place unbiased digits would give about 117; 11
thrice; 44 twice. And the digits alternate between two teams,
4 5 6 8 and 0 1 2 7 9, with 3 belonging to neither: after 73, for
occasion, the following digit is 6, 8, 4 or 5 in 93 p.c of circumstances.

Figure 3. How usually every digit follows one another digit within the public transcription.
The 5 pairs in daring turned out to be the codes for t, p, c, l, and d. The
pale diagonal is the avoidance of doubles. Author’s computation from EHum’s
2016 transcription.

That seems to be like consonants and vowels taking turns. But it doesn’t resolve
whether or not 73 is one code or two frequent neighbors, and no fastened rule of
reducing, akin to pairs at even positions or sure digits all the time starting
a pair, gave a constant end result.

The papal ciphers of the 1540s used a number of designs. In some, one digit
stands for a number of letters and the reader chooses. In some, pairs spell
syllables. One 1545 key writes the vowels as pairs and the consonants as
single digits. ^{I attempted these designs with earlier variations
of the search, with out acquiring readable Italian. The design
that fitted was a easy one additionally printed in Meister’s assortment: each
letter has one code, of 1 or two digits, and the reader tells them aside
by context.}

Figure 4. A key of that design: Meister no. 7, for Alessandro Vitelli,
dated by Meister to 1546. Single digits for some letters, pairs for the remainder,
one null, a couple of phrases. Aloys Meister, Die Geheimschrift (1906),
p. 179;
Getty Research Institute copy, digitized by Internet Archive. Cropped from
the printed version.

The search

I looked for the important thing and the code boundaries collectively. An outer search
modified the assignments of letters to codes. For every candidate key, an
internal decoder looked for a great way to divide the digits and browse them.
The outer search might then examine keys by how nicely their readings scored.

To decide these readings, I educated a personality mannequin that estimates the
chance of every letter from the 4 earlier than it: a five-gram mannequin. It
educated on about 4.9 million letters. Most got here from Machiavelli,
Castiglione, and Vasari on Wikisource. About 720,000 got here from Italian-looking
traces extracted from the OCR of Cardauns’s Nuntiaturberichte aus Deutschland
I.7, which incorporates correspondence of the Farnese secretariat from 1541–44.
That quantity doesn’t print this letter. ^{I lower-cased the textual content,
stripped accents, folded v into u, deleted h, and collapsed doubled
letters. These have been guesses concerning the clerk’s habits, and so they turned out
to match: the letter writes tute, esendo, facia; its key has one code
for u and v and none for h.}

A letter sequence absent from the corpus nonetheless wanted an opportunity. The mannequin
blended the counts for a four-letter context with estimates from shorter
contexts, right down to particular person letter frequencies. This smoothing let it rating
unfamiliar names and broken phrases with out ruling them out altogether.

Reading a candidate key

The decoder walks the digits from left to proper. Under the important thing ultimately
recovered, the opening has these two potential paths, amongst others:

Digits       7 3 4 5 7 4
One path     73  4  57  4    d o p o
Another      7 3 4  57  4    n m o p o

The first path consumes 73 without delay and emits d; the second consumes 7
and 3 individually and emits n m. Each emitted letter provides its base-10 log
chance to the trail’s rating. The letters that comply with can favor one path
over the opposite, so the decoder retains a number of partial readings alive.

Each partial studying is a state, which information the place within the digits,
the final 4 emitted symbols, and the rating up to now. If two paths attain the
identical place with the identical language-model context, their potential
continuations are equivalent: solely the better-scoring path must survive.
Among completely different contexts, the decoder retains the eight greatest at every place.
This is a beam search. It is an approximation: a path discarded early can’t
get well when later letters would have made it convincing.

In the profitable searches, I put aside the digits with a dot or comma above
them and lower the textual content at these locations and at unreadable digits. Each
fragment started with a recent language-model context. The key obtained the
sum of its greatest surviving fragment scores, together with penalties for nulls
and digits it couldn’t decode. The dotted codes would come again as soon as the
letters have been recognized.

Changing the important thing

The outer search started with random assignments, favoring codes that occurred
extra usually. It might swap two assignments, transfer one to an unused code, or
flip off an non-compulsory unit akin to a null. It accepted any enchancment. It
might additionally settle for a worse key, with an opportunity that fell because the run went
on. This is simulated annealing. Accepting worse strikes lets the search
escape a key that beats its instant neighbors however remains to be mistaken. I
repeated the search from recent random keys. Each repetition is a
restart.

The first search to provide recognizable Italian gave this opening, with
underscores marking the null:

dopo_lapartitadelmontepinmcian_il_manon

Dopo la partita del Montepulciano, “after Montepulciano’s departure”. The
cleartext on the identical web page mentions his departure. Further alongside have been
prepararsi contra … del turco, cento naui, ungaria, ridolfo gonzaga,
forteza di luzara, mons. di granuela.

Much of the remainder was nonetheless rubbish. This search had been allowed a pool of
non-compulsory phrase codes, and it used them to clarify away inconvenient digits.
Searches that have been allowed a number of nulls declared three frequent digits
meaningless. Allowing two codes per letter additionally gave excessive scores for worse
textual content. Yet 4 of the primary search’s six restarts had given the identical codes
to the identical eighteen letters, the entire alphabet the clerk used. The letter
assignments have been secure even whereas the additional codes spoiled the studying.

I tightened the search to at least one code per letter, at most one null, and no
phrase codes. I additionally retrained the Italian mannequin with phrase areas eliminated.
A mannequin educated on del quale expects a boundary after the l, however the
decoder presents it with delquale. The unspaced mannequin learns the letter
sequences throughout these boundaries. With these adjustments, the search reached
the identical letter key from two of three restarts. The third stopped at a worse
key.

Table 1. Settings for the tightened search.

Setting within the tightened search	Value
Candidate codes	Single digits and pairs occurring no less than thrice within the scored fragments
Breaks within the enter	203 above-dot/comma tokens and 44 unreadable tokens; one belongs to each teams
Input after these cuts	219 scored fragments; 6,331 retained digits
Beam width	Eight states per digit place
Proposal makes an attempt per restart	500,000
Penalty per null	1; emits no letter and preserves context
Penalty per undecodable digit	4, when neither the one digit nor the out there pair has a code

The penalties discourage a key from enhancing its rating by leaving textual content
unread. Scores beneath divide the penalized complete by the retained-digit rely.
Nearer zero is best. They rank the keys discovered underneath these settings, reasonably
than measuring the chance {that a} studying is appropriate.

Figure 5. The recovered key. The single digits 4, 5, 6, and eight are the vowels
o, a, i, and e that the pairs desk confirmed taking turns with the remainder.
Author’s reconstruction.

Checking the important thing

Could the identical key be recovered with out the opening identify? I cut up the textual content
after its third cipher web page and solved every half by itself, from random
assignments. Every restart recovered the identical eighteen letters and the
null.

I additionally enciphered recognized Italian with two invented keys of the identical design:
one artificial cipher was clear; the opposite had digits substituted or marked
unreadable. Finally, I shuffled the true ciphertext and searched that. These
controls used the tightened search settings.

Table 2. Recovery and management runs.

Input	Result	Score per retained digit
Whole letter	Same eighteen letters and null in two of three restarts	−0.783; third restart −1.150
First and second halves, solved individually	Same letters and null in all twelve restarts, six per half	−0.777 / −0.785
Clean artificial cipher	Letters and null recovered in all three restarts	−0.630
Synthetic cipher with digit substitutions and 44 digits marked unreadable	Letters and null recovered in all three restarts	−0.692
Shuffled actual ciphertext	Best of two restarts	−1.351

The artificial exams didn’t get well the phrase code for et, which the
letter-only search couldn’t specific. Their Italian additionally later proved to
overlap partly with the mannequin’s coaching textual content. All the controls share the
Italian mannequin and its habits, so the test I belief most is the cut up: both
half of the letter alone yields the letter key and null.

Lasry and his coauthors reported that, of their challenge, variable-length
homophonic keys (these permitting a number of codes for a letter) had required
matching plaintext or an already documented key. The less complicated, monoalphabetic
key used right here got here out of the digits.

Dots and the null

With the letters fastened, the decoded textual content round repeated dotted pairs
revealed their meanings. A 27 with the dot on its second digit, the 7,
sat the place Marchese has its che, and wherever the sense needed che,
“that”. A 72 dotted on its second digit too sat the place carichi,
“burdens”, has its chi; a 57, dotted the identical manner, the place non si vede,
“one doesn’t see”, has its non. With the dot on the primary digit as an alternative,
the identical pairs gave qua, que and qui.

Figure 6. The dot on the primary digit provides qua, que, qui; on the second, che, chi,
non. Three extra dotted pairs are titles: Sua Santità, Sua Maestà, Vostra
Signoria, written right here because the cleartext abbreviates them. Author’s
reconstruction.

The search alphabet had additionally included q, which the cipher doesn’t want
as a separate letter. In the tightened full-text run it occupied 81.
The decoded contexts gave that code the studying et, “and”. The null 9
isn’t a constant phrase divider. It ends some phrases and never others, seems
inside phrases (tu·te for tutte, the dot marking the null), and is utilized in
con, written 27 4 9 7. It can stop misreadings: 8 9 0 reads
e s, the place 80 can be learn as t.

What the digits say and what the mannequin needs

Recovering the important thing doesn’t settle the textual content. The sequence 0 5 7 8 0 5
happens 4 instances within the letter. Under the important thing it has two Italian readings.

Figure 7. The digits 0 5 7 8 0 5 underneath the important thing: santa or spesa.
Author’s diagram.

My first studying, in three of the 4 locations, was la santa [impresa], the
holy enterprise, with a phrase equipped. The surrounding argument requires la
spesa, the expense, each time: the Pope will contribute what he can to the
expense; the ships will not be prepared within the 12 months the expense is for; he’ll
pay his share of the expense towards the Turk.

Reading by way of errors

Some passages urged errors within the digits themselves. In Montepulciano’s
identify, the transcription gave 2, which the important thing reads as r, the place the identify
wanted u. I wrote a second decoder to check literal readings with readings
that required small adjustments to the transcription. The key, now together with the
dotted codes, stayed fastened.

The key search had lower the textual content at unreadable and dotted digits. This decoder
labored by way of the cipher on every web page, protecting the language context throughout
unreadable locations. It preserved the transcriber’s {qualifications}:
2? meant a uncertain 2, 1/2 provided two readings, and ? meant an
unreadable digit. Dots, commas, and dashes above the digits have been retained too.

At every place, this system in contrast the following digit or pair with the codes
within the fastened key. An actual match price nothing. Where nothing matched, it
might suggest a special digit or a modified above-mark. It might additionally drop
a digit, provide the lacking half of a pair, or insert a single-digit letter
code. I assigned completely different prices to those operations in order that filling a recognized
gap was simpler than contradicting a digit the transcriber had learn with out
qualification.

Table 3. Edit prices within the fixed-key decoder.

Proposed operation	Cost in base-10 log models
Keep the digit, or the primary provided various	0
Fill an unreadable ?	0.3
Choose the second provided various	0.5
Substitute a listed look-alike for a digit marked unsure	1
Substitute a listed look-alike for a digit not marked unsure	3
Substitute a digit exterior the listing	6
Supply or ignore an above-dot	3
Ignore an above-comma / above-dash	2 / 1
Insert or drop a digit	5

The look-alike listing included such confusions as 1 with 2, 3 with 5
or 7, and 0 with 8 or 9. These have been hand-set prices, not measured
error possibilities.

The rating mixed the 2 sorts of proof:

studying rating = sum of letter log possibilities − sum of edit prices

Code divisions and edits have been searched collectively. A modified digit may be part of
the following one right into a pair, altering each the variety of letters emitted and
the context for the letters after them. As earlier than, the decoder stored a number of
partial readings alive and recorded the alternatives behind each.

At the disputed digit in Montepulciano, protecting 2 emits r with out an
edit price. Choosing 1 emits u however subtracts three from the rating. That
selection additionally adjustments the mannequin’s predictions for the next letters, till
the modified letter has handed out of its four-letter context. The acquire in
the Italian rating was sufficient for u to win. The edit listing information:

Source   Change   Letter   Context
2        2>1      u        elmontep [u] lcian il

This repairs the center of the identify, however its last o remains to be absent.
Another proposal equipped the dot that turns 72, b, into 7̇2, que.
The end result reads in quele di Ungaria, “in these [affairs] of Hungary”, in
place of in ble di Ungaria.

The identical scoring rule might harm a passage that already learn appropriately.
On 71r it proposed studying an 8 because the null 9 and regrouping the close by
digits, turning a le sue sei galere, “his personal six galleys”, into
al tuti galere. I refused that change. The mannequin judged quick letter
sequences. It didn’t know what number of ships the Pope had or comply with the
argument about paying for them. A acquire giant sufficient to pay for an edit
might nonetheless produce the mistaken studying.

The edit listing tied every proposal to its supply digits and surrounding textual content.
I might settle for or reject a studying and find the disputed digits within the
images.

Table 4. Changes proposed by the decoder.

Type of proposed change	Value
Fills of unreadable digits	44
Other proposed operations, together with decisions of an provided various	126
Other operations divided by the general public transcription’s digit rely	1.9%

The totals embody rejected proposals. Whether an obvious error belonged
to the transcriber, the clerk, or the reader nonetheless required checking towards
the web page.

The version

The edited text, the version from right here on, makes use of
sq. brackets for conjectural letters. A
digit-by-digit alignment behind it information what
brackets can’t present: ignored dots, dropped digits, digits learn because the null.
The alignment makes departures seen; it doesn’t make the readings true.

To check the conjectures, I gave the important thing to 2 readers, every a recent run of
the GPT-6 Astra language mannequin, along with the digits of 1 web page decoded
actually underneath it: 72r for one and 71v for the opposite, the 2 pages with the
most departures. Each was requested for a steady studying and an account of
each place the place it left the literal decode. Each recovered its web page’s
argument and most of the version’s phrases, and every caught one thing the
version had mistaken.

On 72r the version had left a niche and equipped [che potrà] on no digits
in any respect. The reader discovered that the twenty digits behind that hole learn verà
la magior piena precisely, “the larger flood will come”. With the misplaced verb
recovered, the complement was pointless. On 71v the version had the Emperor
realizing that the Pope’s forces are small in themselves, S.M.tà s[a] che,
at the price of one 2 learn as 5 and two digits learn as nulls. The
reader’s o[l]tre che, “moreover the truth that”, wants one ignored dot and
one lacking digit, and suits a sentence that goes on to say the forces are
additionally divided. Both have been adopted. The readers have been unbiased of the
version, not of the important thing or the transcription. Like the version, that they had
by no means seen the web page.

The web page

I checked the disputed readings towards images of the manuscript.
In Montepulciano’s identify, the decoder needed a u the place the transcription
gave 2, the code for r. On the web page, the disputed digit is a single
stroke with a foot: the clerk’s 1, which provides u.

Transcription   57  2  03  27
Page            57  1  03  27
Reading          p  u   l   c

Figure 8. F. 70r, second cipher line. AAV, Segr. Stato, Spagna 1A,
by way of DECODE file 92; enlarged element. The identify was already a believable
correction from the Italian. The {photograph} exhibits why the digit was misinterpret.

The clerk usually joins one digit to the following. The high bar of a 7 runs into
the previous digit. The foot of a 1 could make it appear to be a 2. In a small
pilot, I confirmed 4 quick strips to a few imaginative and prescient fashions. They made roughly
12–18 p.c digit errors, usually on those self same confusions. That pilot was
too small, and its scoring too completely different, to determine a character-error-rate
comparability with the general public transcription.

I wanted a option to examine a uncertain digit with the clerk’s different examples.
I wrote a program that first straightened every cipher line and eliminated the
handwriting’s slant. It appeared for cuts within the white gaps and the skinny joins
between digits, then fitted the transcription to these cuts. Each transcribed
digit turned a provisional label on a small picture. A 2 within the transcription
meant “this picture is an instance of a 2”, even the place that label would later
show mistaken.

A seven-neighbor classifier in contrast every picture with examples from different
traces. It used their shapes and proportions to seek out the seven closest
matches, then mixed their labels to suggest a digit. Leaving out the
present line stored a digit from being in contrast with itself or a neighboring
piece of the identical joined stroke.

Figure 9. The glyph transcribed as 2 in Montepulciano, adopted by its seven nearest
matches from different traces. These are the straightened, normalized photos used
for comparability. The labels come from the general public transcription. Author’s
evaluation of AAV, Segr. Stato, Spagna 1A, ff. 70r–73r, images equipped
by way of DECODE file 92.

The classifier agreed with the transcription on 90.5 p.c of the glyphs,
and on 98.9 p.c of the roughly two-thirds of glyphs it learn most
confidently. Those figures measure settlement with the previous labels. To discover
transcription errors, I wanted the disagreements.

For every suspect, a assessment sheet put the digit in its line of handwriting,
beside the seven matches and a reference sheet of the clerk’s types from
0 to 9. I checked the ink and recorded the studying, the rationale, and the way
sure it was. I examined all 196 ranked suspects this fashion, along with
each line the place the segmentation had discovered a special variety of digits
than the transcription. Where the classifier doubted a
digit the version had left alone, my studying largely sided with the
transcriber: a smear might flatten a 3 into the form of a 5, or the
subsequent digit’s stroke might give a 7 the bottom of a 2.

Some errors have been within the variety of digits. On 71v the transcription repeated
the 5 digits that learn anco, “additionally”. The {photograph} has them as soon as.

Transcription   5 7 2 7 4  5 7 2 7 4
Page            5 7 2 7 4
Reading         a n  c  o

Figure 10. F. 71v, fourth cipher line. AAV, Segr. Stato, Spagna 1A,
by way of DECODE file 92; enlarged element. The version had already dropped
the second anco as a repetition by the clerk. The repetition was within the
transcription.

The test discovered 5 such doubled teams, totaling twenty-one digits not
on the web page, in addition to fourteen omitted digits and two swapped pairs. Some
of the additional digits had equipped phrases for the version: sendosi [c]osì firmat[o]
turned sendosi firmat[o], “having been agreed”, with no così, “thus”.

On 70v, a dot and two hooked 1s closed a niche within the account of the naval
bills. The transcription had missed the dot over the 7 and browse each
1s as 2s.

Transcription   27   2 6 9 2 5
Page            27̇  1 6 9 1 5
Reading         che u i · u a

Figure 11. F. 70v, third cipher line. AAV, Segr. Stato, Spagna 1A,
by way of DECODE file 92; enlarged element. With 1 standing for each u and v, this reads
che vi va: the complete phrase is della spesa che vi va dentro, “of the
expense that goes into it”. The dot within the studying row marks the null 9.

Other corrections confirmed letters the version had equipped. In è s[t]ato
on 70r and aiu[t]o on 72r the transcription has 0 8 the place the web page has
80, the code for t. The s of stato nonetheless has no digit behind it: the
web page has the null 9 there, so the version now reads è [s]tato. Other
bracketed letters, ris[p]osta and sol[d]o amongst them, stand on the web page
because the version learn them. Some letters the version had equipped prove to
be on the web page: the n of Massimi[li]a[no] is a 7 the transcription
learn as 9; the c of ues[c]ou[o], vescovo, bishop, is a 2 the
transcription skipped, finishing 27; the i of Dor[ia] is a 6
dropped at a line finish.

On 72r the version had conte[n]ti with a equipped n. The web page has a 7
the place the transcription has 4, and contenti is actual. On 70v the
conjecture [ca]pitarà il bisogno is withdrawn: the transcription’s 8 is a
1, and the version now reads [d]’ovunche più sarà il bisogno, “wherever the
want will likely be larger”. The preliminary d stays conjectural. Other checks went
the opposite manner. On 73r the version had [ne]l qual luogo by altering a 6,
however the 6 is apparent and the textual content reads il qual luogo. On 70r the 2 the
version learn as 3 for contra [l]’armata has the clerk’s typical 2 type:
if l was meant, the clerk miswrote it.

One correction modified who had requested for a public announcement. Under the
archive stamp on 72r, the corrected digits give la S.M.tà [h]a voluto,
“His Majesty has wished”, the place the version had learn S.S.tà, His Holiness.
The Emperor had requested for the Hungarian subsidy to be provided publicly at
Speyer, within the Pope’s identify.

The fort’s identify stays conjectural. Where the version had equipped
[Poviglio] from the Gonzaga family tree, the corrected digits give
a pore in · pui ·, with the dots right here marking nulls. The version now reads
a porre in P[o]ui[glio], “to settle in Poviglio”. The lacking letters nonetheless
come from the proposed historic identification.

I additionally checked the dots wherever the version added or ignored one. In
risposta, the mark transcribed above a 7 was the highest bar of the 5
earlier than it. Other obvious dots belonged to strokes from the road above. In
Marchese, the dot was actual however on the mistaken digit of the pair: the clerk
had written qua the place the identify wants che.

The corrections judged positive or possible went into a brand new transcription.
Unresolved readings stayed open. I then aligned the version towards these
corrected digits.

Table 5. Source digits consumed by the version’s alignment. Both the
transcription and the edited studying modified after the web page test.

How the version reads the digits	Before the web page test	After
Digits within the transcription	6,577	6,562
Read precisely underneath the important thing, together with nulls	6,216	6,358
Read with a recorded departure	300	170
Left in gaps	61	34

Blots, the archive stamp, and smears alongside the web page edges go away some digits
and dots unread. I’ve not checked each glyph the classifier accepted,
and substitutions between not like digits nonetheless want one other look. The day
of the month additionally stays unread.

A up to date copy in clear might assist with the remaining readings. The
Archivio Apostolico Vaticano searched the index to a register cited by
Cardauns, however discovered no entry for an April 1542 letter from Farnese to
Poggio.

What Farnese put in cipher

The abstract beneath follows the proposed decipherment. Its quotations
translate the edited Italian. Gaps and conjectures stay within the version.

After Montepulciano’s departure, the imperial ambassador in Rome, the
Marqués de Aguilar, has been with the Pope and introduced him Andrea Doria’s
recommendation on arming towards the Turkish fleet, and a request for assist.

The Pope will maintain again nothing inside his means, at sea or in Hungary,
wherever the necessity seems to be larger. But no person but is aware of whether or not the
larger hazard will come by land or by sea, and he is not going to resolve the place to
spend, “in order to not fall into the error of spending in useless in a single place
that wanted it much less, after which be unable to assist the opposite.” He leaves that
to the Emperor’s prudence.

He has two doubts concerning the plan to arm 100 ships. He argues {that a}
combined fleet of galleys and crusing ships has by no means performed nicely. And 100
ships would take so lengthy to suit out that they’d hardly serve this 12 months,
the 12 months the cash is for. He can pay his share anyway, past his personal six
galleys.

The troopers on his ships should be his personal. The Papal State is stuffed with good
troopers used to going to struggle. Unless he retains them employed, no edict in
the globe will cease them taking pay from whoever provides it, to the
disturbance of his personal state and no benefit to the Emperor. He means, too,
to remain impartial in order that he can dealer the peace, and to offer no person grounds
to say he pays for something however protection towards the Turk.

Then there are the opposite commitments. The Hungarian subsidy, which Charles V
requested for at Lucca, was settled in Rome with the imperial minister Granvelle.
At the Emperor’s request, the bishop of Modena, Giovanni Morone, has now
provided it publicly, within the Pope’s identify, on the Diet of Speyer, the meeting
of the German estates then in session. The enclosed extract from Modena’s
letter exhibits that these on the Diet are usually not content material and ask for far more.
Ancona, Civitavecchia, the opposite ports, and the guard of Lombardy additionally want
cash. The Pope can’t but say what extra he’ll give on both
aspect.

Finally, Ridolfo Gonzaga, son of Gianfrancesco, is alleged to have taken a
fee from somebody whose identify remains to be unread. He gathered infantry at
Luzzara, his brother Massimiliano’s fort in Mantuan territory, then moved
to his personal fort within the territory of Parma and commenced to fortify it, both
as a result of the Marchese del Vasto, the governor of Milan, was making ready to
take away him by pressure or for some design of his personal. Whether that is mere
rashness on Ridolfo’s half or one thing bigger isn’t but clear. The
Pope’s legate has orders to cease him, by pressure if essential, and troops
have moved into Parma. Vederassi il successo: we will see the way it
goes.

Morone’s surviving studies independently verify a public presentation of
papal assist at Speyer on 23 March and complaints that the supply was too small.
They corroborate the decipherment’s account, though the actual
enclosed extract has not been recognized. ^{The Pope and the Emperor
had met at Lucca in September 1541, two weeks after the Turks took
Buda.}

The Gonzaga family tree additionally suits. Gianfrancesco’s sons included Massimiliano,
who held Luzzara, and Rodolfo, spelled Ridolfo within the letter, related
with Poviglio within the Parma territory. A printed archival stock
describes a 1546 letter to Rodolfo as “signore di Poviglio”. That helps
the proposed fort identify with out supplying its lacking letters.

The peace Montepulciano was carrying didn’t come. Francis I declared struggle on
Charles V in July 1542. By the top of that month Montepulciano was again in
Italy. On 30 July the nuncio Girolamo Verallo reported Ferdinand’s assertion
that la pratica della tempo period exclusa: the negotiation for peace was
completed.

Status of the decipherment

This is a candidate studying. The secret’s secure: both half of the letter
alone yields the identical letters and null. The web page test examined the disputed
readings and the classifier’s suspects, not each glyph.

No different cryptanalyst has but verified or challenged the answer. On
15 September I despatched the important thing and plaintext to MysteryTwister, whose group
checks Level X options by hand.

Sources and analysis supplies

The notes establish the proof for particular person claims, together with the
article’s personal computations and proposed readings. References to printed
books use their printed web page numbers; manuscript references use folios.
Cipher-line numbers rely solely traces containing cipher. English translations
comply with the cited Italian passages. The version provides phrase division,
punctuation, and a few spelling; its brackets and alignment file unsure
readings. Online catalog and problem statuses have been checked on
16 September 2026.

Manuscript and transcription

• Archivio Apostolico Vaticano (AAV), Segr. Stato, Spagna 1A,
  ff. 70r–73v. Letter recognized right here as Alessandro Farnese to Giovanni
  Poggio, Rome, April 1542; day unresolved.
  DECODE record 92,
  identifier ASV/i1025/SdS/Spain/IA/2. Older references use ASV,
  Archivio Segreto Vaticano. The April date is the studying of this letter’s
  dateline, reasonably than the broader date vary assigned to the archival bundle.
• EHum. Transcription of ASV_i1025_SdS_Spain_IA-2, 9 January 2016.
  Public original;
  complete transcription, including the transcriber’s header;
  ciphertext appendix and notation. This is the enter to
  the important thing search. The
  manuscript-corrected transcription
  is a later analysis product, used for the version’s last alignment.

Published sources

• Borrelli, Luciano, ed. “Biblioteca trentina [parte seconda].”
  Studi trentini di scienze storiche, Sezione prima, 68, no. 2 (1989):
  185–214. Digitized article.
• Cardauns, Ludwig, ed. Berichte vom Regensburger und Speierer
  Reichstag 1541, 1542. Nuntiaturen Verallos und Poggios. Sendungen
  Farneses und Sfondratos 1541–1544. Nuntiaturberichte aus Deutschland
  nebst ergänzenden Aktenstücken, I. Abteilung, vol. 7. Berlin, 1912.
  Digitized edition.
  Cited beneath as Cardauns, with doc quantity and printed web page.
• Cascella Alcaraz, Sara. “Una cifra imperial en Roma (1543).”
  Estudios Románicos 34 (2025): 63–73.
  doi:10.6018/ER.613381.
• Fabbrici, Gabriele. “Geografia politica della pianura reggiana fra
  XVI e XVII secolo. Note introduttive.” In Corti e diplomazia
  nell’Europa del Seicento: Correggio e Ottavio Bolognesi (1580–1646),
  edited by Blythe Alice Raviola, 51–66. Mantova: Universitas Studiorum, 2014. Cited page 61.
• Gairdner, James, and R. H. Brodie, eds. Letters and Papers,
  Foreign and Domestic, of the Reign of Henry VIII. Vol. XVII, 1542.
  London: Her Majesty’s Stationery Office, 1900.
  British History Online, pp. 286–300.
• Gayangos, Pascual de, ed. Calendar of State Papers, Spain.
  Vol. 6, half 2, 1542–1543. London: Her Majesty’s Stationery Office, 1895. British History Online, pp. 1–13.
• Lämmer, Hugo, ed. Monumenta Vaticana historiam ecclesiasticam
  saeculi XVI illustrantia: ex tabulariis Sanctae Sedis apostolicae
  secretis. Freiburg im Breisgau: Herder, 1861.
  Digitized edition.
• Lasry, George. “The Vatican Challenge — Part 5.” MysteryTwister,
  7 July 2019.
  Challenge page;
  challenge PDF.
• Lasry, George, Beáta Megyesi, and Nils Kopal. “Deciphering papal
  ciphers from the sixteenth to the 18th Century.” Cryptologia 45, no. 6
  (2021): 479–540. First revealed on-line 23 June 2020.
  doi:10.1080/01611194.2020.1755915.
• Litta, Pompeo. “Gonzaga di Mantova,” tavola XVI, “Marchesi di
  Luzzara estinti nel 1794.” In Famiglie celebri di Italia. Milano,
  sheet dated 1835. BnF/Gallica, view 26.
• Luzio, Alessandro. L’Archivio Gonzaga di Mantova. Vol. II,
  La corrispondenza familiare, amministrativa e diplomatica dei Gonzaga.
  Verona: Officine grafiche A. Mondadori, 1922.
  Digitized edition.
• Megyesi, Beáta, Nils Blomqvist, and Eva Pettersson. “The DECODE
  Database: Collection of Historical Ciphers and Keys.” In Proceedings
  of the 2nd International Conference on Historical Cryptology (2019),
  69–78. Open proceedings PDF.
• Meister, Aloys. Die Geheimschrift im Dienste der päpstlichen Kurie
  von ihren Anfängen bis zum Ende des XVI. Jahrhunderts. Paderborn:
  Ferdinand Schöningh, 1906.
  Digitized edition.
• Pastor, Ludwig von. Geschichte der Päpste seit dem Ausgang des
  Mittelalters. Vol. V, Paul III. (1534–1549). Freiburg im Breisgau:
  Herder, 1909. Digitized edition.
  Dates and pages beneath check with this 1909 version.

Individual entries within the Dizionario Biografico degli Italiani are cited
by their named authors, volumes, years, and on-line entry titles within the notes.
Their on-line textual content doesn’t provide printed web page breaks. Methodological
references and the signed institutional historical past of Buda are likewise given
in full on the related notes.

Research knowledge and illustrations

The edition appendices comprise the key tables,
machine-readable key,
edited Italian text, and
digit-by-digit alignment. The
automatic reading preserves the decoder’s proposals,
together with these rejected within the version. The
transcription appendix consists of the general public and
corrected texts and the person manuscript choices. The
results supplement studies the search
and management outcomes, mannequin particulars, and unbiased rereadings. The
earlier alignment provides the before-check
figures in desk 5. These are the article’s analysis outputs, not an
independently transmitted plaintext.

The figure source register information supply photos, folios,
crop coordinates, transformations, and determine knowledge. The public previews and the
full-resolution manuscript images are completely different information; the latter have been
equipped by way of DECODE with restricted entry and are usually not reproduced in full
right here. The reproduced particulars and processed glyphs are recognized individually
within the register.

Notes